Feedback That Moves in All Directions

Category: Safety Enables Signal Accuracy improves when feedback is lateral and upward, not only hierarchical.

On the bridge of a ship approaching a narrow harbour entrance, the captain calls a heading. He has run this passage many times, and the number he calls is the one that has always worked.

A junior officer, two years out of her training and the least senior person standing watch, has been tracking the same approach on her own plot. The figures do not quite agree. The tide is running harder than the captain’s heading assumes, and on her reckoning the ship will be set a little too far toward the eastern bank before the turn. It is not a dramatic gap. It is the kind of small discrepancy that is usually nothing, that a tired person on a familiar approach would round away without a second thought.

On a great many bridges, she would say nothing. The captain has decades on her. He has called this heading a hundred times. To question it is to risk being wrong in front of everyone, or to risk being right and resented for it, and either way the safe career move is to assume the experienced man has seen something she has not and keep her plot to herself.

On this bridge, she says it. “Captain — I’m reading the set a touch stronger than that. I think we’re being carried east.” She says it plainly, without apology and without drama, because that is simply how things are reported here. The captain does not bristle. He looks at her plot, looks at his own, and adjusts two degrees. The turn comes off cleanly. Afterwards he says, in front of the watch, that she was right to call it, and means it.

Nobody on that bridge thinks anything remarkable has happened. That is the point. The most junior person corrected the most senior, in the moment, on his own ground, and the ship was better for it — and the only reason it was possible is that someone, long before, had decided it must be.

The Principle

Feedback that travels only top-down gives you a confident picture and a false one. Accuracy arrives when feedback also flows sideways, between peers, and upward, from the junior to the senior — because the people closest to the raw signal are usually not the people at the top, and a channel that only runs downward never carries what they can see.

The default assumption is that feedback is something seniors give to juniors. The boss reviews the work, the experienced hand corrects the new one, instruction flows down the gradient of authority because that is where the expertise is presumed to live. There is real truth in that, and it is not the whole truth. Some of the most decision-relevant information in any environment sits at the bottom and the edges — in the person reading the plot, watching the queue, handling the part nobody senior touches that day. If feedback only runs downhill, that information has nowhere to go. The strongest environments treat feedback as something that must move in every direction at once, and they understand that the lateral and upward directions are the hard ones, the ones that do not happen on their own, the ones that have to be built.

Why It Is Inevitable

This is not a refinement the best environments add for elegance. It is forced on them, because the alternative quietly starves them of the information they most need.

The first reason is simply where knowledge sits. The person with the authority to decide is rarely the person with the clearest view of the facts. Seniority buys experience and breadth, but it costs proximity. The further up you go, the more you see and the less of it you see directly — you see summaries, headlines, the version that survived the climb. The detail that actually determines whether a decision is right lives with the people doing the work, and they are, almost by definition, junior to the people deciding. An environment that only lets feedback flow downward has arranged things so that its best information can never reach its most consequential choices.

The second reason is that error is no respecter of rank. Experienced, senior, well-regarded people are wrong sometimes — not through any failure of character, but because they are working from a familiar assumption that has quietly stopped being true, or because they are tired, or because the one fact that matters this time happens to be the one they cannot see from where they stand. If the only people permitted to correct them are even more senior, most of their errors will go uncorrected, because most of the people positioned to catch them are below them. The correction has to be allowed to travel upward, or it does not travel at all.

And the third reason is that peers see things sideways that no one sees up or down. Two people doing similar work, at the same level, notice each other’s blind spots in a way a manager never will, because they are close enough to the same task to recognise the small wrongness in it. Lateral feedback catches a whole class of error that hierarchical feedback structurally misses. So any environment that wants a true picture is pushed toward the same conclusion: it needs all three directions running, because each one carries signal the others cannot.

How It Shows Up

• A junior person can correct a senior person in the moment, on the senior person’s own ground, and the senior person takes it as information rather than as challenge.
• Peers routinely check and sharpen each other’s work without a manager in the loop, and nobody treats it as going behind anyone’s back.
• Bad news travels upward as readily as good news, and arrives while it is still small — not suddenly and fully formed, but early and tentative, as the first faint reading.
• The phrase that gets used is closer to “I’m seeing it differently” than “you’re wrong,” and it costs the person saying it nothing.
• Seniors visibly thank people for upward corrections, including ones that turn out to be unnecessary, because the willingness to call it is the thing being protected.
• The most useful information in the room is not assumed to live at the top of it; people ask the person closest to the work, not the person highest in the structure.

Why It Causes Benefit

When feedback moves in all directions, an environment gets something it cannot get any other way: its decisions are made on real information rather than on the comfortable, smoothed version that authority alone produces.

The benefit is, first, simple accuracy. The picture the deciders work from is closer to reality, because the people who can actually see reality have a route to put what they see in front of the people who act on it. Errors are caught regardless of who made them — a senior person’s wrong assumption is corrected by the junior who can see it is wrong, a peer’s oversight is caught by the peer beside them — and they are caught early, while they are still small drifts and not yet crises. The environment is not relying on its most senior people being right, which is fortunate, because no one is always right. It has arranged things so that being wrong gets noticed and said.

The deeper benefit is what it does to the flow of bad news. In an environment where feedback only runs downhill, bad news is dangerous to carry upward, so it is held, softened, and quietly dropped at every level — and the people who could act on it learn of it last and least. When upward feedback is safe and normal, that whole failure simply does not occur. The faint early signal — the plot that does not quite agree, the number that has drifted, the thing that is probably nothing — reaches the people who can act while it is still cheap to act on. The environment can see its own mistakes clearly, from the bottom and the edges as well as the top, which is the precondition for correcting them. That is why these places do not merely avoid the occasional disaster; they keep getting more accurate over time, because nothing true is being kept from the people who decide.

How To Cultivate It

• Make upward correction safe by modelling it from the top. The first time a senior person visibly thanks a junior for catching their error — and means it, in front of others — does more for the culture than any amount of stated policy. The moment a senior person punishes an upward correction, even subtly, the channel closes.
• Separate the correction from the rank, out loud, until it is understood. A junior reading the plot differently is not insubordination and not a status contest; it is information about reality. Saying so plainly, and repeatedly, is how the two stop being confused.
• Build lateral feedback into the work itself, so peers checking peers is the normal way things get finished rather than something someone has to decide to do. Feedback that is routine carries no accusation; feedback that only happens when someone is worried always will.
• Ask for the upward signal by name, especially the bad and the uncertain: “what are you seeing that I’m not?” The people closest to the work will not always volunteer it across a rank gap — the gap is real even when the safety is — so the senior person has to reach across and pull it.
• Reward the early, tentative call even when it turns out to be nothing. If an unnecessary warning costs the person who raised it, you are training everyone to wait until they are certain, and certainty about a problem usually arrives too late to help.
• Watch the response, not just the rule. People read what actually happens to the person who corrects upward far more than they read what they are told is allowed. One bad reaction, witnessed, teaches more than a year of encouragement.

What Good Looks Like

An environment where feedback runs freely in every direction and nobody finds any direction strange. Where the most junior person can correct the most senior in the moment, peers sharpen each other’s work as a matter of course, and bad news climbs upward as easily as good news — all of it understood as information about reality rather than as challenge to anyone’s standing, so it costs no one anything to give or to receive. Where the people with authority assume that the clearest view of the facts probably sits below them and at the edges, and arrange things so that view can reach them. The faint early signal gets heard while it is still faint, because the person holding it has a safe route to say it and a senior person glad to listen. The picture the deciders work from is rough and true rather than smooth and false, and decisions are made on what is actually happening. It looks, from outside, like a place where rank matters less than it should — but what is really happening is that rank has been stopped from distorting the information, so that the right call can be made by whoever can see it, whatever their place in the structure.

A Reflective Question

In your environment, when did someone junior to you last correct you to your face — and if you cannot easily remember it happening, is that because you are rarely wrong, or because the channel that would carry the correction upward was never built?